This study was done to evaluate whether there were any differences in microleakage of class V composite restorations according to restoration site and cavity size.

Total sixty-four restorations were made in molar teeth using Esthet-X. Small (2 × 2 × 1.5 mm) and large (4 × 2 × 1.5 mm) restorations were made at the buccal/lingual surface and the proximal surface each. After 1,000 times of thermocycling (5℃ - 55℃), resin replica was made and the percentage of marginal gap to the whole periphery of the restoration was estimated from SEM evaluation.

Thermocycled tooth was dye penetrated with 50% silver nitrate solution. After imbedding in an auto-curing resin, it was serially ground with a thickness of 0.25 mm. Volumetric microleakage was estimated after reconstructing three dimensionally.

Two-way ANOVA and independent T-test for dye volume, Mann-Whitney U test for the percentage of marginal gap, Spearman's rho test for the relationship between two techniques were used.

The results were as follows:

1. The site and size of the restoration affected on the microleakage of restoration. Namely, much more leakage was seen in the proximal and the large restorations rather than the buccal/lingual and the small restorations.

2. Close relationship was found between two techniques (Correlation coefficient = 0.614 / P = 0.000).

Within the limits of this study, it was noted that proximal and the large restorations leaked more than buccal/lingual and the small restorations. Therefore, it should be strictly recommended large exposure of margins should be avoided by reducing unnecessary tooth reduction.